A Contemporary Introduction to Natural Language Processing(NLP) in 2020
The art of making machines understand human language.
This article is written to serve the purpose of a short 5 min read to introduce new learners and beginners with the awesome world of NLP. It article touches upon the following basic aspects of NLP
I. What is NLP?
II. Why is it difficult to work with Textual data?
III.Real-World Applications of NLP.
I. What is NLP?
Language is an important part of human communication.
Read the following sentence carefully :
“ A rich and creamy whole-wheat pasta dish filled layer by layer with refreshingly fresh onions and garlic, lathered in a succulent sauce and topped with imported, premium quality mozzarella.”
The human brain helps us visualize this lasagna dish after reading its description on a hotel menu. But the computer doesn’t know what is the context of the above sentence or what a lasagna is?
So to make the computer understand the language(words, sentences, paragraphs) and make sense of this textual data, we use NLP.
Formally, Natural Language Processing is the application of computational techniques for the analysis and the synthesis of text. NLP aims to give computers the ability to do tasks involving human language.
II. Why is it difficult to work with Text?
Comprehending language is tough for computers. Listed below are a few of the common challenges in NLP :
1.Synonymy —
As in different words having the same meaning.
Sentence 1: “The President of the United States has signed a new decree”
Sentence 2: “POTUS has inked in a new law”
Both these statements are advocating the same sentiment. However as they are completely different sentences syntactically, computers have a hard time figuring out the user intent.
2. Lexical Ambiguity —
Meaning a lack of clarity of the same word being used in different contexts.
S1: “The bank deposit rate is quite high”
S2: “He stood near the bank admiring the river”.
Here, the word “bank” has 2 different meanings. In the first case it represents a financial institution, whereas in the second case it refers to land near the river. The computer gets confused as to which bank we are referring to.
Consider the example too :
“She killed the man with the tie”
Here we ask,
Was the man wearing the tie? OR,
Was the tie the murder weapon?
3. Language related issues —
Every language has its uniqueness.
In English we do not have a gender associated with the noun ‘bread’. But in the Russian language, each noun is assigned a gender (masculine, feminine, or neuter )And bread is ‘Masculine’ !!
The grammar and morphology of languages are so different and unique making it tough for machines to understand the language.
Know more about Russian grammar here
4.Referential Ambiguity -
S1: “Jeff is my friend. He likes to sell”.
In the second statement “he” refers to Jeff. It is difficult for the computers to decipher what person/entity the pronoun he is referring to.
5.Out of Vocabulary problem -
Machines can only handle data that they have seen before. It is unable to adapt well to new words. (Unsupervised problem)
6.Language generation -
For chatbots to work effectively, they need to first understand the constructed sentences which are grammatically correct, understand its meaning and context and effectively reply to customer queries.
After reading all these issues while working with texts we realize NLP is hard. But further, let us take a look at some of the interesting use cases of NLP!
NLP can be cool too!!!
III. Real-World Applications of NLP
Here are some of the applications we come across as a part of our daily life.
1.Speech Recognition:
Amazon’s Alexa, Apple’s Siri, etc. are the voice assistants used to convert voice commands, translating them to perform computer tasks like playing some music.
Also, Google’s ‘OK Google’ feature used as a trigger word for waking Google’s Assistant and home devices
2.Sentiment Analysis:
Sentiment analysis is the interpretation and classification of common emotion of the target audience & finding if the text is leaning towards a positive, negative or neutral sentiment using text analytics.
It is immensely useful for Businesses in figuring the overall sentiment of their products (Amazon reviews), movies (IMDb ratings), food and restaurants (Zomato reviews), etc.
Tweets gone wrong!!
Tesla boss Elon Musk wiped $14bn (£11bn) off the carmaker’s value after tweeting its share price was too high. Link
3.Machine Translation :
Machine Translation involves the conversion of textual data from one language to another.
Eg.Spanish to English Translation(Google’s Translate)
4.Conversational AI :
This includes the exchange of text chats with a machine and get basic queries resolved with the help of chatbots.
Example: WHO launches a chatbot on Facebook Messenger to combat COVID-19 misinformation. Link
5.Topic modeling:
This method involves detecting word and phrase patterns within paragraphs/textual documents, and automatically clustering word groups and similar expressions that best characterize that set of documents.
Application: Auto-tagging of customer queries in Customer Service to ensure quicker response and faster solution.
6.Parts of Speech Tagging :
POS tagging helps to find out the various nouns, adverbs, verbs, and map them in a sentence.
Application: Word sense disambiguation. In the sentences “I left the room” and “Left of the room”, the word left conveys different meanings. A POS tagger would help to differentiate between the two meanings of the word left.
7.Named Entity Recognition(NER):
Identifying a person’s name, time, location, date, and other entities present in text. It is heavily used in the field of, medicine, bioinformatics, and molecular biology.
8.Text Summarization :
It is a method used to compress a text document, creating a summary of major points of documents.
Its applications include News summary(Inshorts app), Novel/Book Summary (Blinkist) etc.
9.Speech to Text /Text to speech conversion.
A popular product to better explain Text to Speech would be Amazon’s Audible that reads out texts in the ebooks.
10.Text Prediction /Autocorrection
This application is seen in products like Gmail auto-mail suggestion. Also Grammarly uses NLP to solve grammatical errors and syntactical errors.
Thank you for reading all the way through! I hope this article helped you get an idea of this wide field of NLP and sparked an interest in this space. I tried to keep it brief, simple, and interesting. This is my first Medium publication and would appreciate any feedback.
Let me know your thoughts in the comments section below. Feel free to reach out to me on LinkedIn.
